EO-based environmental impact assessment of camps: a differentiated spatial view using multi-scale satellite imagery

[Jamon Van Den Hoek]

Dear Colleagues,

Our third paper is now posted: S. Lang (2021). EO-based environmental impact assessment of camps: a differentiated spatial view using multi-scale satellite imagery. https://www.populationenvironmentresearch.org/pern_files/statements/EO-based%20environmental%20impact%20assessment%20of%20camps_Lang.pdf

In this paper, the author draws on their years of experience in providing environmental remote sensing capabilities in refugee and IDP settings. The paper examines the potential of satellite imagery for both demand-driven detection of refugee and IDP settlement dwellings as well as more long-term monitoring of land cover condition and change. In walking us through the considerations and routes for translating raw satellite data to semantic, analysis-ready information, the paper provides a synoptic view of state-of-the-art Earth Observation techniques for humanitarian applications.

Please have a read and share your thoughts and questions here.

Thank you,

Jamon

--

Jamon Van Den Hoek, PhD | Asst Professor

Conflict Ecology Lab | @JamonVDH

Geography Program, CEOAS | Oregon State University

Oregon State University occupies land in the traditional territory of the Ampinefu ("Mary's River") band of the Kalapuya. After the Kalapuya Treaty (Treaty of Dayton) in 1855, Kalapuya people were forcibly removed to what are now the Grand Ronde and Siletz reservations, and are now members of Confederated Tribes of the Grand Ronde Community of Oregon and the Confederated Tribes of Siletz Indians.

[Jamon Van Den Hoek]

Hello PERN community,

Thank you, Stefan, for sharing your work today! 

First, I was wondering if you might comment on how short-term on-demand/rapid response humanitarian-oriented remote sensing complements longer- term land cover/land use change monitoring in refugee and IDP settings. In your experience working with humanitarian NGOs in refugee and IDP settings, under what circumstances and at what point does the interest or need shift away from on-demand detection towards a longer-term monitoring of change?

Does this transition occur -- if it occurs at all -- when the expectation of temporariness or ephemeral habitation gives way to the likelihood of protraction or long-term settlement?

Second, in light of Andrew Kruczkiewicz's paper from yesterday on anticipatory humanitarian decision-making, how might we as a research-practitioner community better integrate on-demand and long-term remote sensing analysis towards anticipatory action?

How should we collect data for anticipatory action in complex environments when the complexity itself is so novel and also dynamic?

Thanks for the contributions and discussions so far,

Jamon

[Lang Stefan]

Thanks Jamon for starting this discourse!

 

First and foremost our population estimation routines serve the immediate needs and the related humanitarian supply chain in emergency phase. This refers to planning and coordination of logistics, of medical services and health care and water / sanitation infrastructure is of utmost importance. Here, (close-to) real-time monitoring is the appropriate means, in the highest detail possible.

 

In the longer term (care and maintenance phase including recovery/repatriation), the observation and assessment of impacts becomes more critical, but the scale of observation might gradually get coarser. Guiding questions are: How concrete population figures relate to the carrying capacity of the impacted spaces, how do resources diminish, to which degree the settlement structure is self-sustaining, what are secondary effects to security, but also increasing risk to health (e.g., Malaria) due to shifting land use patterns. 

 

And here it transitions to the second part of your question, touching upon the issue of to which degree EO technology can be used to actually 'predict'. We often use the term spatial prediction, when referring to the interpolation of singular observations to space. Remote sensing, one may think, does not require this step, due to its spatially exhaustive sampling technique. However what me measure is not the phenomena itself, but - at best - one indication to it. We have an area intensive variable which needs to be translated, interpreted. And here we have to types of challenges, spatial resolution (the sampling distance) and the semantic gap, e.g. what means a reflection value of XY in terms of degradation? So spatial prediction means the translation of observations to information.

 

Now, (how) can we move from spatial prediction to temporal prediction (i.e., forecast)? Well, as compared to weather forecast we are not quite there yet. Systemic behaviour might be fairly predictable if only enough variables are modelled but this might be (still) limited for human group behaviour. Here data assimilation comes into play. We are experimenting with combining EO data with digital activities traced from social media data or any other real-time response of human presence. I did not highlight this in the paper ... but it is of course another quite important step in data assimilation. Anticipatory action requires the understanding of trends, of likely behaviour, of predictable patterns. Here we are at the dawn of what hybrid AI may provide in the future, while extrapolating trends in the biophysical domain (such as soil degradation, human-induced droughts, flooding etc.) is surely more straight forward then the actions of  humans.

 

Happy to discuss this further, the exciting part is definitely what future capacities emerge from this convergence  of different strands of technology coincide, once domain / practical expertise is fully integrated and real added-value information products are achieved.

 

All the best,

Stefan

--
The Population-Environment Research Network (PERN) Cyberseminar Discussion List
---
You received this message because you are subscribed to the Google Groups "PERNSeminars - List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pernseminars...@ciesin.columbia.edu.
To view this discussion on the web visit https://groups.google.com/a/ciesin.columbia.edu/d/msgid/pernseminars/CAHasiqdx%2Be07FcNoMZdZw%3DUHb%3DF5CaVOaLF11EbHEbH7JW5DUw%40mail.gmail.com.